Approximately half of the U.S. population will experience a significant mental health problem during their lifetime, including 29% with anxiety pathology severe enough to qualify for an anxiety disorder diagnosis. Critically, the majority will not receive treatment, creating a serious need to consider alternative approaches to delivering mental health services that can meet needs on a larger scale. Cognitive Bias Modification (CBM) interventions for anxiety hold considerable promise as a way to meet these needs. These computer-based training programs are designed to alter biased ways of thinking, such as selective assignment of threat interpretations, which are known to cause and maintain anxiety. CBM programs have shown efficacy in the laboratory, and now have the potential to be tested and implemented on a broader scale using a web- based infrastructure. The current proposal aims to develop a web-based CBM infrastructure to train interpretations, and to evaluate and maximize the usability, acceptability, and feasibility of web-based CBM for anxiety symptoms. In Aim 1, a web-based interpretation bias training program will be built using the Project Implicit Mental Health (PIMH) infrastructure, an existing website directed by the Principal Investigator. This approach encourages efficiencies by capitalizing on the existing site and its heavy traffic. The training program will be piloted on a small test group of anxious participants (N=15) and an advisory board (N=8) of anxiety researchers, clinicians, and experts in CBM to obtain feedback on the programs' usability and acceptability. Aims 2 and 3 will evaluate target engagement, feasibility, and effectiveness of the web-based CBM program among individuals with moderate to severe anxiety symptoms, and will assess the inclusion of an anxious imagery prime to enhance CBM's effectiveness on the web. Participants will be randomly assigned to either 8 sessions of active CBM (100% positive scenario training) or a 50% positive/50% negative standard control condition, or a neutral, non-valenced control condition. Half the participants in each of these 3 conditions will receive an anxious imagery prime prior to each training session, and half will receive a neutral imagery prime, resulting in a 3 training condition x 2 prime design (N=300; target of n=50 per condition). Feasibility will be determined by analyses of recruitment, attrition, acceptance of randomization, adherence to and appropriateness of the measurement model, caseness, extent of missing data, and safety. Additionally, target engagement (change in interpretation bias), mechanisms underlying the prime's effects, and preliminary tests of effectiveness at reducing anxiety symptoms will be evaluated via assessments at baseline, and following sessions 3, 6, and 8, and at a 2-month follow-up. This research fits directly within the priorities of NIMH, given the proposal's focus on evaluating feasibility and effectiveness of a psychosocial intervention for a prevalent and impairing mental health problem that has strong potential for dissemination to large, diverse populations.